Animal models
Table of Contents
The C57BL/6 mouse line was used for all of the experiments (The Jackson Laboratory and National Institutes of Aging). Homozygous Pf4-KO mice were previously generated and characterized as described48. Pf4-KO mice were a gift from M. Anna Kowalska. Heterozygous mice were bred to generate Pf4-KO and WT littermate controls. Cxcr3-deficient mice (Cxcr3-KO) were created and characterized by Deltagen. Homozygous Cxcr3-KO mice were acquired from The Jackson Laboratory (005796). Breeding mates consisting of heterozygous females and hemizygous males were established to produce experimental male Cxcr3-KO and WT controls, and female Cxcr3-KO and heterozygous controls. All other studies were performed with male mice, unless indicated otherwise. The numbers of mice used to result in statistically significant differences was calculated using standard power calculations with α = 0.05 and a power of 0.8. We used an online tool (http://www.stat.uiowa.edu/~rlenth/Power/index.html) to calculate power and sample size on the basis of experience with the respective tests, variability of the assays and interindividual differences within groups. Animals used for each individual experiment were from the same vendor and aged together. All animals from Jackson Laboratories were acquired at 2 months of age and were aged in-house. The mice were moved to a new location for behavioural assessment. The two locations where we conducted behavioural analysis are on the UCSF Parnassus campus: the UCSF Rehabilitation Behaviour Core and the Villeda Lab Behavioural Suite. Mice were housed under specific-pathogen-free conditions under a 12 h–12 h light–dark cycle, with humidity maintained at 30–70% and temperature at 68–79 °F (20–26 °C). All animal handling and use was performed in accordance with institutional guidelines approved by the University of California San Francisco IACUC.
Animal plasma collection, platelet fraction preparation and systemic administration
Mouse blood was collected by intracardial bleed at time of euthanasia from young (3 months old) and aged (20 months old) mice. Blood was collected with EDTA, followed by centrifugation at 1,000g for plasma preparation. For western blot and ELISA analysis, plasma was aliquoted and stored at −80 °C until use. Before systemic administration, plasma was dialysed using 3.5 kDa D-tube dialyzers (EMD Millipore) in PBS to remove EDTA. For platelet fraction preparation, plasma was centrifuged at 20,000g, the supernatant was discarded and the pelleted platelet component was resuspended in an equivalent volume of saline. Aged mice were systemically treated with saline, plasma or the platelet fraction (100 μl per injection) through intravenous tail vein injection 8 times over 24 days. Likewise, saline or PF4 (5 μg ml−1) was systemically administered to mice (100 μl per injection) through intravenous tail vein injection 8 times over 24 days. Recombinant mouse PF4 (CHM-245, ProSpec) and human-platelet-derived PF4 (CHM-234, ProSpec) were dissolved in sterile ultrapure water at 100 µg ml−1 (according to the manufacturer’s instructions), before final dilution in saline.
Tissue collection
Mice were anesthetized with 87.5 mg per kg ketamine and 12.5 mg per kg xylazine and transcardially perfused with ice-cold phosphate-buffered saline. Tissues were removed and processed for subsequent analysis. To process the brains, either the hippocampus was subdissected and snap-frozen or the whole brain was fixed in phosphate-buffered 4% paraformaldehyde, pH 7.4 at 4 °C for 48 h before cryoprotection with 30% sucrose.
RNA isolation and bulk RNA-seq
RNA was isolated from either the whole hippocampus or microglia isolated from the hippocampus. Microglia were isolated from hippocampi by enzymatic and mechanical dissociation using the Neural Tissue Dissociation Kit (P) (Miltenyi Biotec) according to the manufacturer’s instructions. Myelin was depleted using Myelin Removal Beads (Miltenyi Biotec). Subsequently, microglia were captured using CD11b microbeads (Miltenyi Biotec). Total RNA was isolated from samples by lysis using TRIzol Reagent (Thermo Fisher Scientific), separation with chloroform and precipitation with isopropyl alcohol, according to the manufacturer’s instructions. After RNA isolation, RNA-seq libraries were constructed using the Smart-Seq2 protocol49 with modifications. In brief, 1 ng high-quality RNA was reverse-transcribed using SuperScript II (Life Technologies, 18064-014) with a poly-dT anchored oligonucleotide primer, and a template switching oligonucleotide primer that generated homotypic PCR primer binding sites. The cDNA underwent 10 rounds of PCR amplification using KAPA HiFi Hotstart (Kapa Biosystems, KK2601), followed by Ampure bead (Agencourt) clean-up. The quality of the amplified cDNA was tested by qPCR analysis of Gapdh and nucleic acid quantification. A total of 1 ng of high-quality amplified cDNA was fragmented with the Tn5 transposase from the Illumina Nextera kit (FC-131-1096) to a median size of around 500 bp. The fragmented library was amplified with indexed Nextera adapters (FC-131-1002) using 12 rounds of PCR. The final libraries were purified with Ampure beads and quantified using the qPCR Library Quantification Kit (Kapa Biosystems, KK4824). Libraries were pooled for sequencing on the Illumina HiSEq 2500 system (paired-end reads 2 × 100 bp). Alignment of RNA-seq reads to the mouse mm10 transcriptome was performed using STAR (v.2.7.3a)50 using the ENCODE standard options, read counts were generated using RSEM (v.1.3.1) and differential expression analysis was performed in R (v.3.6.1) using the DESeq2 package (v.1.38.0)51 (detailed pipeline v.2.0.1 and options are available at GitHub (https://github.com/emc2cube/Bioinformatics/)). Genes significantly changed after treatment with both plasma and the platelet fraction were determined using a nominal P < 0.05, and significance in microglia after systemic PF4 administration was determined with a nominal P < 0.01. GO term enrichment analysis was performed using Enrichr52 (GO Biological Process 2018). Heat maps were generated using Morpheus (https://software.broadinstitute.org/morpheus/).
RT–qPCR
To quantify mRNA expression levels, equal amounts of cDNA were synthesized using the High-Capacity cDNA Reverse Transcription kit (Thermo Fisher Scientific, 4368813), then mixed with SYBR Fast mix (Kapa Biosystems) and primers. Gapdh was amplified as an internal control. RT–qPCR was performed in the CFX384 Real Time System (Bio-Rad). Each sample and primer set were run in triplicates and relative expression levels were calculated using the \({2}^{-\Delta \Delta {C}_{{\rm{t}}}}\) method53.
Immunohistochemistry
Tissue processing and immunohistochemistry was performed on free-floating sections according to standard published techniques26. Cryoprotected brains were sectioned coronally at 40 μm with a cryomicrotome (Leica Camera). Free-floating coronal sections (40 μm) were incubated overnight with anti-IBA1 (1:1,000, Wako, 0191741; or 1:1,000, Synaptic Systems, 234-004), rat anti-CD68 (FA-11, 1:250, Bio-Rad MCA1957), rabbit anti-C1q (1:500, Abcam ab182451) and anti-phosphorylated-CREB (Ser133) (1:2,500, Millipore 06-519) primary antibodies. Labelling was revealed using secondary antibodies (donkey anti-rabbit conjugated Alexa Fluor 555 (1:750, Life Technologies, A31572), donkey anti-rat conjugated Alexa Fluor 647PLUS (1:750, Invitrogen, A48272) donkey anti-guinea pig conjugated Alexa Fluor 488 (1:750, Jackson ImmunoResearch, 706-545-148) and goat anti-rabbit, biotinylated (1:500, Vector, BA-1000)). Labelling of biotinylated antibodies was revealed using the Vectastain Elite ABC-HRP Detection Kit (Vector, PK-6100) with diaminobenzidine (Sigma-Aldrich, D5905). Sections were imaged using confocal microscopy (Zeiss LSM800 or Zeiss LSM900) or bright-field microscopy (Keyance). Individual cell numbers and intensity in the dentate gyrus was quantified across 3–4 sections per animal using ImageJ.
Collection and preparation of human platelet-rich plasma
Blood was collected from healthy young men (aged 20–35 years) or healthy older men (aged 60–75 years), who volunteered for either a cross-sectional or non-randomized single-arm study at the UCSF Human Performance Center. Samples used in this experiment were from the cross-sectional baseline timepoint only. The participants were asked to complete detailed surveys about their health habits, including physical activity, and donate a fasting blood sample. This study was approved by the Institutional Review Board of UCSF. Samples were drawn from consenting participants by the UCSF Clinical and Translational Science Institute Blood laboratory, and immediately transported to the Villeda laboratory for processing. One 6 ml aliquot of blood was centrifuged at 500g for 8 min (4 °C) and the plasma was collected. The plasma was aliquoted and centrifuged again at 700g for 17 min (12 °C). Platelet-poor plasma was removed as the top 70% of the solution and the remaining solution was used to resuspend the pellet as platelet-rich plasma. The samples were aliquoted and stored at −80 °C until use.
Western blot analysis and ELISA
For Western blot analysis, samples were combined with RIPA lysis buffer (Abcam, ab156034) with complete protease inhibitor (4693116001, Sigma-Aldrich) and phosphatase inhibitor (Thermo Fisher Scientific, 78420). Subsequently, the samples were mixed with 4× NuPage LDS loading buffer (Invitrogen, NP0008), loaded onto an SDS polyacrylamide gel (Invitrogen) and transferred onto a nitrocellulose membrane. Equal loading of samples was confirmed using Ponceau S solution (Sigma-Aldrich, P7170) and membranes were imaged with the ChemiDoc System (Bio-Rad). The blots were blocked in 5% milk in Tris-buffered saline with Tween-20 and incubated with anti-GAPDH (6C5, 1:5,000, Abcam, ab8245) goat anti-mPF4 (1 µg ml−1, R&D Systems, AF595), mouse anti-hPF4 (170138, 0.5 µg ml−1, R&D Systems, MAB7952), mouse anti-thrombospondin-1 (A6.1, 1:200, Santa Cruz, sc-59887, C2519) or rabbit anti-cyclophilin A (1:200, ENZO Life Sciences, BML-SA296-0100). Horseradish-peroxidase-conjugated secondary antibodies (donkey anti-goat conjugated HRP (1:2,000, Invitrogen, A15999), goat anti-mouse conjugated HRP (1:2,000, Millipore, AP124P), and donkey anti-rabbit conjugated HRP (1:2,000, GE Healthcare, NA934V)) and an ECL kit (GE Healthcare) were used to detect protein signals. Developed membranes were imaged using the ChemiDoc System (Bio-Rad). Selected images were exported and quantified using ImageJ (v.2.0.0). Plasma protein levels were measured by ELISA for PF4 (Mouse CXCL4/PF4 Quantikine ELISA Kit; R&D Systems; MCX400), CCL2 (Mouse CCL2/JE/MCP-1 Quantikine ELISA Kit; R&D Systems; MJE00B), TNF (mouse TNF Quantikine ELISA Kit; R&D Systems; MTA00B) and β2-microglobulin (Cloud-Clone Corp; SEA260Mu) according to the manufacturer’s protocol.
HDTVI of HiBiT plasmids
To detect the localization of HiBiT-tagged PF4 and TRF in various mouse tissues, mice were hydrodynamically injected with GFP, PF4-HiBiT or TRF-HiBiT constructs as previously described7,54. To generate plasmids, RNA was isolated from mouse peripheral blood mononuclear cells (PBMCs) or liver using TRIzol reagent (Thermo Fisher Scientific, 15596026) and the PureLink RNA Mini Kit. RNA was reverse-transcribed using the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, 4368813) and oligo dT primers (Promega, C1101). The following primers were used for PCR amplification of the Pf4 coding sequences and partial 3′ and 5′ untranslated regions from the PBMC cDNA library: forward, CACCAGTGGCACCCTCTTGACAT; and reverse, GGCAGCTGATACCTAACTCTCC. Trf was amplified from liver cDNA using the following primers: forward, CACCAGCGGGTCGGTCTGTACTC; and reverse, CAGTGGCAACCCACCTCTTG. Pf4 and Trf ORFs were cloned into the pENTR D-TOPO vector (Thermo Fisher Scientific, K240020) and sequence verified using M13F and M13R sequencing primers. Restriction enzyme sites (Nhel and EcoR1 for Pf4 and Nhel and Mfel for Trf), a Kozak sequence and a C-terminal HiBiT tag were added during an additional PCR amplification step. The resulting PCR fragments were cloned into a mammalian expression plasmid using the designated restriction sites. The bicistronic plasmid vectors expressed Pf4 or Trf and an IRES eGFP reporter using a CMV promoter. An empty IRES eGFP construct based on the same plasmid was used as a control. All coding plasmid sequences were verified by Sanger sequencing. Endotoxin-free plasmid kits were used for plasmid preparation before in vivo use. To perform HDTVI of constructs, plasmid DNA (50 μg) was suspended in 3 ml saline and injected in the tail vein in 5–7 s in mice. At 24 h after HDTVI, the mice were euthanized and plasma was collected by intracardial bleed. After perfusion, the hippocampus, cortex, cerebellum and liver were dissected, snap-frozen and lysed in RIPA lysis buffer (Abcam, ab156034) with cOmplete protease inhibitor (Sigma-Aldrich, 4693116001) and phosphatase inhibitor (Thermo Fisher Scientific, 78420). A total of 20 μg of protein from each sample was loaded in duplicate in an opaque 96-well plate (Corning, 353296). HiBiT luminescence was measured on the Cytation 5 (BioTek) using the Nano-Glo HiBiT Lytic Detection System (Promega, N3030) according to the manufacturer’s instructions.
Splenocyte isolation
Splenocytes were isolated from young saline-treated, aged saline-treated and aged PF4-treated mice. For collection of splenocytes, spleens were removed, mechanically dissociated with a syringe plunger over a 70 μm cell strainer and washed with 10 ml of ice-old RPMI medium with 2% FBS. Cells were centrifuged and RBC lysis was performed (155 mM NH4Cl, 1 mM KHCO3 and 0.1 mM EDTA). Subsequently, cells were washed and resuspended in staining buffer.
CITE-seq
CITE-seq analysis was performed on splenocytes isolated from five mice per group. Cells from all five mice from each group were pooled for CITE-seq antibody labelling, as previously described55 (detailed methods are available online (https://citeseq.files.wordpress.com/2019/02/cite-seq_190213.pdf)). In brief, the samples were blocked with TruStain fcX (BioLegend) for 10 min on ice. After the blocking step, the samples were incubated on ice with Total-seqB antibodies purchased from BioLegend (Extended Data Table 1). After 30 min of labelling, samples were washed three times and filtered through a 40 µm cell strainer before delivering the prepared samples to the UCSF-IHG Genomics Core for analysis with the 10x Genomics Chromium Single Cell Expression Solution 3′ kit with Feature Barcode Technology (v.3.1). The Genomics Core prepared cells for 10x Genomics Chromium single-cell capture. 10,000 cells were loaded per sample. cDNA libraries were prepared according to the standard 10x Genomics protocols. The final library pool was sequenced to a depth of 30,000 cDNA reads per cell and 3,000 ADT reads per cell on the NovaSeq 6000 S2 system. The raw base sequence calls were demultiplexed into sample-specific cDNA and ADT files with bcl2fastq/mkfastq sample sheet using Cell Ranger (10x Genomics). CITE-seq analysis and statistical analysis of raw FASTQ files were processed using the Cell Ranger software package (10x Genomics) for the RNA expression matrix and CITE antibody counts matrix. The data were combined using the Cell Ranger aggrpipeline (10x Genomics). Downstream single-cell analysis was performed using the R package Seurat56. Data were processed to remove doublets and unwanted sources of variation by removing cells with more than 5,000 and fewer than 300 genes per cell and regressing on number of UMIs. Genes expressed in fewer than three cells were filtered out. Cells with a percentage of mitochondrial genes of higher than 10% were removed. The matrices of data were log-normalized in a sparse data matrix and PCA was applied to reduce dimensionality. The first 20 PCA components were used to cluster cells by Louvain clustering implemented in Seurat while UMAP plots were independently generated to aid in 2D representation of multidimensional data independent of the clustering. Log-normalized gene expression data were used for visualizations with violin plots, UMAP plots and generation of heat maps.
Flow cytometry
Flow cytometry analysis of platelets was performed on whole blood and the platelet fraction of the plasma preparation. After collection, blood was diluted 1:10 in 250 mM EDTA. Platelets were labelled with anti-CD61 PE (2C9.G2, HMβ3-1, 1:50, BioLegend, 104308) at 4 °C. Cells were washed and resuspended in PBS for analysis with the BD LSR II Flow Cytometer. Flow cytometry analysis of splenocytes was performed, as previously described57. Cells were labelled with two independent panels to assess specific populations of myeloid cells and lymphocytes. Antibodies for the myeloid panel were as follows: CD45-BUV395 (30-F11, 1:200, BD, 564279), CD3-APC (17A2, 1:200, Tonbo Biosciences, 20-0032-U025), B220-APC (RA3-6B2, 1:200, BioLegend, 103211), CD49b-APC (DX5, 1:200, eBioscience, 50-112-9698), Ly6G-BV711 (1A8, 1:200, BioLegend, 127643), I-A/I-E Alexa Fluor 700 (M5/114.15.2, 1:200, BioLegend, 107621), F4/80-PeCy7 (BM8, 1:200, eBioscience, 25480182) and CD11b-BV650 (M1/70, 1:200, BioLegend, 101239). Antibodies for the lymphoid panel were as follows: CD45-BV711 (30-F11, 1:200, BD, 563709), B220-PeCy5 (RA3-6B2, 1:200, eBioscience, 15-0452-82), CD4-PeCy7 (RM4-5, 1:200, eBioscience, 25-0042-82), CD3-APC (17A2, 1:200, Tonbo Biosciences, 20-0032-U025), CD8a-Pacific Blue (5H10, 1:200, Thermo Fisher Scientific, MCD0828), CD62L-PerCP-Cyanine5.5 (MEL-14, 1:100, Tonbo Bioscience, 65-0621-U100) and CD44-APC eFluor 780 (IM7, 1:100, eBioscience, 47-0441-82). In brief, 1 × 106 cells were blocked with FBS and stained at 4 °C. Thereafter, the cells were washed, fixed with 4% paraformaldehyde solution, washed and resuspended in PBS for storage until analysis using the BD LSR II Flow Cytometer.
T cell culture and activation
For in vitro experiments, untouched T cells were isolated from aged mouse spleens using the Pan T Cell Isolation Kit II, mouse (Miltenyi Biotec), according to manufacturer’s instructions. Isolated splenocytes were resuspended in staining buffer (PBS, pH 7.2, 0.5% BSA and 2 mM EDTA) before addition of biotin-conjugated antibody. After a 5 min incubation at 4 °C, additional buffer and anti-biotin microbeads were added to the solution. Cells were incubated for 10 min and the suspension was applied to a washed LS column on a QuadroMACS Separator (Miltenyi Biotec), while the unlabelled flow-through was collected as the pan T cell fraction. Cells were washed and plated in culture medium (RPMI with 10% FBS and 55 μM BME). Subsequently, T cells were stimulated (Dynabeads Mouse T-Activator CD3/CD28 for T-Cell Expansion and Activation; Thermo Fisher Scientific) with or without recombinant mouse PF4 (1 μg ml−1; Prospec). Cells were incubated for 3 days before collection for analysis. Flow cytometry analysis of the cells was performed using the following antibodies: CD3-eFluor 660 (17A2, 1:100, eBioscience, 50-0032-82), CD4-PE-Cyanine7 (RM4-5, 1:200, eBioscience, 25-0042-82), CD8a-PE (53-6.7, 1:100, BioLegend, 100708) and CD279/PD-1-FITC (29F.1A12, 1:200, BioLegend, 135214). In brief, 1 × 106 cells were blocked with FBS and stained at 4 °C. Thereafter, the cells were fixed with 4% paraformaldehyde solution, washed and resuspended in PBS for storage until analysis with the flow cytometer (BD Accuri C6 Plus).
NOR
The NOR task was performed as previously described7. On day one (the habituation phase), mice performed open field testing by exploring an empty arena for 10 min. Infrared photobeam breaks were recorded and movement metrics were analysed using the MotorMonitor software (Kinder Scientific). On day two (the training phase), two identical objects were placed into the habituated arena, and the mice were allowed to explore for 5 min. On day three (the testing phase), one object was replaced with a novel object, and the mice were allowed to explore for 5 min. The time spent exploring each object was quantified using the Smart Video Tracking Software (Panlab; Harvard Apparatus). Two different sets of objects were used. To control for any inherent object preference, half of the mice were exposed to object A as their novel object and half to object B. To control for any potential object-independent location preference, the location of the novel object relative to the trained object was also counterbalanced. To determine the percentage of time with the novel object, we calculate (time with novel object)/(time with trained object + time with novel object) × 100. Mice that did not explore both objects during the training phase were excluded from the analysis.
Y maze
The Y Maze task was conducted as previously described7. During the training phase, the mice were placed into the start arm facing the wall and were allowed to explore the start and trained arm for 5 min, while the entry to the 3rd arm (novel arm) was blocked. The maze was cleaned between each mouse to remove odour cues, and the trained arm was alternated between mice. After training, the mouse was returned to its home cage. After 45 min, the mouse was returned to the start arm and was allowed to explore all three arms for 5 min. The number of entries and the time spent in each arm was quantified using the Smart Video Tracking Software (Panlab; Harvard Apparatus). The percentage of entries in each arm was defined as the number of entries in each arm divided by the total number of entries in all arms during the first minute of the task. The discrimination index was quantified by (novel arm − trained arm)/(novel arm + trained arm). Mice that did not perform three entries during the first minute of testing were excluded.
Fear conditioning
In this task, mice learned to associate the environmental context (fear conditioning chamber) with an aversive stimulus (mild foot shock; unconditioned stimulus) enabling testing for hippocampal-dependent contextual fear conditioning. To assess amygdala-dependent cued fear conditioning, the mild foot shock was paired with a light and tone cue (conditioned stimulus). Freezing behaviour was used as a readout of conditioned fear. Specific training parameters were as follows: tone duration of 30 s; level of 70 dB, 2 kHz; shock duration of 2 s; intensity of 0.6 mA. This intensity is not painful and can easily be tolerated but will generate an unpleasant feeling. On the training day (day 1), each mouse was placed in a fear-conditioning chamber and was allowed to explore for 2 min, during which time freezing was recorded to assess the baseline freezing behaviour. Subsequently, a 30 s tone (70 dB) and light, ending with a 2 s foot shock (0.6 mA) were delivered. Then, 2 min later, a second unconditioned-stimulus–conditioned-stimulus pair was delivered. On the testing day (day 2), each mouse was first placed into the fear-conditioning chamber containing the same context, but with no CS or foot shock. Freezing was recorded for 2 min. Then, 1 h later, the mice were placed into a new context containing a different odour, cleaning solution, floor texture, chamber walls and shape. The animals could explore for 2 min before being re-exposed to the conditioned stimulus. Freezing was analysed for 1–3 min using a FreezeScan video tracking system and software (Cleversys).
RAWM
Spatial learning and memory were assessed using the RAWM paradigm, according to an established protocol58. In this task, the mouse was trained to the location of a constant goal arm throughout the training and testing phase. The start arm changed each trial. Entry into an incorrect arm was scored as an error, and errors were averaged over training blocks (three consecutive trials). During training (day 1), the mice were trained for 12 trials (blocks 1–4), with trials alternating between a visible and hidden platform. After an hour break, learning was tested for 3 trials (block 5) using only a hidden platform. During testing (day 2), the mice were tested for 15 trials (blocks 6–10) with a hidden platform. When scoring, investigators were blinded to treatment.
Data, statistical analyses and reproducibility
All experiments were randomized and blinded by an independent researcher before tail vein injection. Researchers remained blinded throughout histological, biochemical and behavioural assessments. Groups were unblinded at the end of each experiment on statistical analysis. Data are expressed as mean ± s.e.m. The distribution of data in each set of experiments was tested for normality using the D’Agostino–Pearson omnibus test or Shapiro-Wilk test. Statistical analysis was performed using Prism v.8.0 or v.9.0 (GraphPad). Means between two groups were compared using two-tailed unpaired Student’s t-tests. Comparisons of means from multiple groups with each other were analysed using one-way ANOVA followed by the appropriate post hoc test, as indicated in the figure legends. Trial by group interactions were analysed using repeated-measures ANOVA with Šidák’s correction for multiple comparisons. Additional statistical details are indicated in the respective figure legends. All data generated or analysed in this study are included in this Article. The main experimental findings are representative of two independently performed experiments. All replication attempts were successful. RNA-seq and CITE-seq data were not replicated due to resource limitations, but were orthogonally validated. Experimental replication was not attempted for negative data.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.